In 1993, Vernor Vinge wrote of a coming technological singularity where artificial super intelligence (ASI) ignites explosive growth in knowledge creation.
So, where will all this knowledge come from?
Knowledge creation is a multifaceted process that drives innovation and progress. It can be viewed through three progressive steps:
- Compiling existing knowledge
- Synthesizing latent knowledge
- Discovering original knowledge
How will AI accelerate knowledge creation at each step?
Compiling — Building the Vantage Point
Isaac Newton expanded our understanding of the natural world with his laws of motion, gravity, and (to the chagrin of math students) the invention of calculus.
To his credit, Newton modestly acknowledged his debt to compiled knowledge in a letter to Robert Hooke in 1676:
“If I have seen further, it is by standing on the shoulders of giants.”
Newton recognized his breakthroughs were built upon discoveries by ‘giants’ like Euclid, Copernicus, and Descartes.
Beyond those luminaries, we should also thank the nameless army of scribes who built the vantage point from which Newton could see “further.”
Does anyone remember who planted that apple tree?
AI Large Language Models (LLMs) similarly owe their abilities to the broad shoulders of the public internet’s digital human content.
Notably, ChatGPT 3 was trained on more than three times the volume of the entire US Library of Congress.
Think of your chatbot as a librarian who’s read every book.
But, AI isn’t just a bookworm. It absorbs knowledge from multiple mediums like video, music, conversations— even whale songs.
Unlike libraries or the internet, AI doesn’t use card catalogs or keyword searches. AI compiles what it has learned into a knowledge network, enabling it to draw upon the connected whole when answering questions.
To top it off, natural language processing (NLP) has given the vast trove of human knowledge a voice — transforming archives into dialogs where users can explore and learn.
I wonder what questions we will ask?
Synthesis — Connecting the Dots
Greek philosopher and scientist, Aristotle recognized the power of connected knowledge with his declaration:
“The whole is greater than the sum of its parts.”
That sage phrase captures the essence of synthesis — uniting individual elements into a system with capabilities beyond those of its separate components.
By combining knowledge, perspectives, and data, we unlock emergent properties that don’t appear when each piece stands alone.
Indeed, synthesis accounts for just about everything we create.
For a practical example, consider how automobiles combine disparate knowledge of metallurgy, chemistry, aerodynamics, physics, and electronics. The same holds for the cornucopia of products we all enjoy.
Synthesis is the powerhouse wringing value from compiled knowledge — and it’s about to get an AI upgrade.
One of AI’s amazing powers is its ability to connect the dots of knowledge in almost infinite combinations. What used to take months of painstaking research now takes minutes — sometimes revealing unseen possibilities.
For example, Google’s AlphaFold AI, used 60 years of compiled knowledge of how amino acids connect to predict the 3D structures of 200 million proteins —the building blocks of life.
Among that multitude of structures lay secrets of diseases and cures, resilient crops and ocean health, new materials, and industrial catalysts. Where will the possibilities end? Perhaps the fountain of youth?
It is essential to recognize that synthesis does not produce knowledge directly. Instead, it poses a hypothesis (Ho) to be tested through experiments and observation.
Even the inventors of AlphaFold would say their protein predictions are only hypotheses. Proving them true or false is the job of the next stage of knowledge creation.
Discovery – Proving and Producing
The renowned inventor Thomas Edison famously proportioned the effort in making knowledge breakthroughs in his proclamation:
“Genius is 1% inspiration and 99% perspiration.
The statement emphasizes that true discovery and invention demand constant experimentation and the willingness to learn from mistakes.
Edison spoke from experience — he tried and failed 2,774 times before finding a design for the electric light bulb. Someone hand that man a towel.
We also see that an original hypothesis like “Can we use electricity to heat a metal wire and make light?” yields thousands of sub cycles, “How thick a wire? Which metal? What voltage?”
Inspiration may come from connecting the dots or serendipitous observation, but the sweaty work of proof and product inevitably follows.
How will AI assist in reducing the 99% perspiration problem? Three key areas will drive acceleration:
Tighter Hypotheses: AI fine-tunes hypotheses by analyzing massive datasets, identifying patterns, and suggesting the optimum experimental and prototyping paths, reducing the number of dead ends.
Faster Cycles: AI accelerates iterative testing by automating data analysis and optimizing experiments in real time. Such AI-assisted drug discovery could reduce years of trial and error to months.
Physical Agency: AI-integrated robotic labs will perform the physical labor, enabling 24/7 experimentation and exponentially increasing the speed of discovery.
As information technology strategist Paul Clermont explains:
“What puts the S in ASI is not just how much it knows going in, more than any individual or whole lab could know, but how quickly it can use it to synthesize and discover new knowledge.”
Creating A Trusted Partner
How do we steer Artificial Super Intelligence to become a partner of humankind rather than its master?
Put these on the to-do list:
Build a Solid Foundation: Disinformation, deep fakes, and recycled AI content threaten the integrity of AI’s knowledge foundation. Rigorous data curation and adherence to scientific principles are vital to preserve AI’s integrity.
Merge the Cognitive and Physical: Current generative AI models lack real-world experiential knowledge. This gap will partially close as cognitive AI merges with physical AI into a unified compilation stream.
Guard Safety and Privacy: With great power comes great responsibility. Ironclad guardrails must be employed to prevent AI from disseminating dangerous methods and private information.
Assure Open Access: Proprietary AI can lead to inequitable concentrations of knowledge. Society must enable equal opportunity and propel progress by appending Open AI Access provisions to existing declarations.
Enforce Transparency: The potential for AI to manipulate our behavior grows as it becomes increasingly entrenched in our lives. We must redouble efforts to insist on transparency into influencing activities.
Mind The Environment: The resources needed to power and cool AI exceeds those of small nations. We cannot let the fear of being left behind blind us to the fact that ASI will be useless if we don’t have a habitable planet.
Beware Anthropomorphism: As AI interacts with humans, it will become adept at mimicking human behavior and emotions, but without a lived experience, AI will lack true emotional intelligence.
The Fuse Has Been Lit
It took nature 13.8 billion years of natural selection to create the human brain. It took humans less than a century to create AI.
At that pace, we have little time to get ASI right.
Ensuring that ASI becomes a trusted ally requires a collective commitment from researchers, policymakers, and society.
Ethical frameworks must evolve alongside AI capabilities, balancing innovation and resource use with accountability.
Public discourse and governance must prioritize inclusivity, ensuring that AI’s benefits are distributed rather than concentrated among a privileged few.
By fostering a culture of transparency, responsibility, and shared progress, we can guide ASI’s development toward enhancing humanity’s potential rather than diminishing it.
